Have you ever noticed how DBAs, SAs and developers alike seem to get very attached to their databases in an almost emotional sense? I worked on a feature last month with a DBA who detailed a strategic plan for how to keep Oracle out of the workplace. I suppose it’s like anything in tech really, if I have to step outside of OS X Leopard for long I start to feel troubled and slightly dizzy.
Of course that’s just how us individuals feel. In the real world there’s not just Oracle (did someone say ‘thank god for that’?), there’s also MySQL, Sybase, MS SQL Server and DB2-UDB environments (and we could go on…) to consider. We know that in practice, any medium to large sized data shop will probably have a heady blend of more than one of the aforementioned.
The recipe for success in these data melting pots is often argued to be heterogeneous surveillance and monitoring tools that work in real time. I wonder if my anti-Oracle chum would have been mollified with one of these systems in place? So I decided to dig a little deeper.
Software products are available that act as real-time monitoring tools capable of overseeing an ‘entire’ database environment. Typically, they have drill-down capabilities for problem identification and are built to allow DBAs to view performance data without writing scripts.
First on my personal Google-o-meter for this type of technology was Bradmark. I mailed my pal about these guys and he was mildly placated, but still not convinced.
I told him about BMC and Quest as well – they’re comparable products in some senses – but he still wasn’t too happy. Bradmark positions its ‘Surveillance’ product as one complete out of the box tool while BMC and Quest, it appears to me, require two products to offer the same functionality.
I told him that products like this are sold on the basis of ‘allegedly’ offering unattended event management to monitor database environments when the DBA is not available – and at that point I got some interest. So I asked for some third party opinion.
“I think it’s interesting that there is still a place for such tools, despite (or perhaps because) of the increasing complexity of RDBMS over the past decades. There’s always more that can be done – but the question is, how compelling are such tools to the people that need to use them? Its difficult to read much from the rhetoric, but the trick with such tools is to support the DBAs and developers, enhancing their role and not to automate them away. Which would neither be desirable nor practical,” said analyst Jon Collins of Freeform Dynamics.
In an attempt to be as even handed as possible, I inspected the typical offering a little more closely. Bradmark says its Surveillance product has, “A ‘cluster window’ function for a views into availability and load across cluster nodes, views of up and down instances (alerts on down instances), view of logical cluster topology and review of historical cluster actions.”
There’s also a window view into the status of interconnects and another to show the performance of cluster caches. Rounding out my investigation I also found newly redesigned support for Microsoft SQL Server 2008 – once again for heterogeneous database monitoring – and MySQL support slated for Q3.
So which is it then, automation good or automation bad? Well, given that the subject matter in hand is heterogeneous database environments, one might logically argue that a bit of automation (if you can coral and control it) is always a good thing. Especially if it shoulders the burden of writing scripts that can cause uncertainty and confusion if project members leave.
So was my pal the DBA any happier when I sent him this research/blog. Actually he and his wife had just had a baby, so he would probably opt for as much automation as he can get right now.
Perhaps that’s it then, if we look at tools in the heterogeneous data environment, we need to look at the WHOLE environment and the big picture before we can make any rational plans.